ARTICLE IN PRESS
Atmospheric Environment 41 (2007) 7837–7850
www.elsevier.com/locate/atmosenv
Atmospheric concentrations and dry deposition fluxes of
particulate trace metals in Salvador, Bahia, Brazil
Pedro A. de P. Pereiraa,b, Wilson A. Lopesa, Luiz S. Carvalhoa,
Gisele O. da Rochab,1, Nei de Carvalho Bahiaa, Josiane Loyolac,
Simone L. Quiterioc, Viviane Escaleirad, Graciela Arbillac,
Jailson B. de Andradea,b,
a
Instituto de Quı´mica, Universidade Federal da Bahia, Campus de Ondina, 40170-115 Salvador, BA, Brazil
Centro Interdisciplinar de Energia e Ambiente—CIEnAm, Campus do Canela, UFBA, 40110-040 Salvador, BA, Brazil
c
Departamento de Fı´sico-Quı´mica, Universidade Federal do Rio de Janeiro, CT, Prédio A, Sala 408, 21949-900 Cidade Universitária,
Rio de Janeiro, RJ, Brazil
d
Centro Nacional da Pesquisa do Solo, EMBRAPA, Rua Jardim Botânico 1024, 22460-000 Rio de Janeiro, RJ, Brazil
b
Received 13 October 2006; received in revised form 8 June 2007; accepted 8 June 2007
Abstract
Respiratory system is the major route of entry for airborne particulates, being the effect on the human organism
dependent on chemical composition of the particles, exposure time and individual susceptibility. Airborne particulate trace
metals are considered to represent a health hazard since they may be absorbed into human lung tissues during breathing.
Fossil fuel and wood combustion, as well as waste incineration and industrial processes, are the main anthropic sources of
metals to the atmosphere. In urban areas, vehicular emissions—and dust resuspension associated to road traffic—become
the most important manmade source.
This work investigated the atmospheric concentrations of TSP, PM10 and elements such as iron, manganese, copper and
zinc, from three different sites around Salvador Region (Bahia, Brazil), namely: (i) Lapa Bus Station, strongly impacted by
heavy-duty diesel vehicles; (ii) Aratu harbor, impacted by an intense movement of goods, including metal ores and
concentrates and near industrial centers and; (iii) Bananeira Village located on Maré Island, a non-vehicle-influenced site,
with activities such as handcraft work and fishery, although placed near the port. Results have pointed out that TSP
concentrations ranged between 16.9 (Bananeira) and 354.0 mg m3 (Aratu#1), while for PM10 they ranged between 30.9
and 393.0 mg m3, both in the Lapa Bus Station. Iron was the major element in both Lapa Station and Aratu (#1 and #2),
with average concentrations in the PM10 samples of 148.9, 79.6 and 205.0 ng m3, respectively. Zinc, on the other hand,
was predominant in samples from Bananeira, with an average concentration of 145.0 ng m3 in TSP samples, since no
PM10 sample was taken from this site. The main sources of iron in the Lapa Station and Aratu harbor were, respectively,
soil resuspension by buses and discharge of solid granaries, as fertilizers and metal ores. On the other hand, zinc and
copper in the bus station were mainly from anthropic contributions, probably heavy-duty vehicle ageing and wearing
actions releasing off Zn from tires and Cu from brake linings. In the Aratu harbor, the high copper concentrations found
were probably due to the port’s activities, as discharges of copper concentrate on its terminal, although other sources could
Corresponding author. Instituto de Quı́mica, Universidade Federal da Bahia, Campus de Ondina, 40170-115 Salvador, BA, Brazil.
Tel./fax: +55 71 32375524.
E-mail address: [email protected] (J.B. de Andrade).
1
Current address: Instituto Multidisciplinar em Saúde, Universidade Federal da Bahia, 45055-090 Vitória da Conquista, BA, Brazil.
1352-2310/$ - see front matter r 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.atmosenv.2007.06.013
ARTICLE IN PRESS
7838
P.A. de P. Pereira et al. / Atmospheric Environment 41 (2007) 7837–7850
be possible, as burning of diesel fuel on ships and heavy oil in heaters. Finally, the Bananeira site has been presented as a
different profile, since this remote site has shown zinc as the most abundant element, demonstrating to have an unexpected
anthropic contribution. On a mass-to-mass basis, both zinc and manganese were in high levels in the Bananeira site and
their presence strongly suggest the impact of other sources, such as the Industrial Center of Aratu and/or a siderurgy plant,
not far away from that location.
r 2007 Elsevier Ltd. All rights reserved.
Keywords: Trace metals; ICP OES; PM10; TSP; Bus station; Harbor and island; Brazil; Salvador
1. Introduction
Respiratory system is the major route of entry for
airborne particulates, being the effect on the human
organism dependent on chemical composition of the
particles, exposure time and individual susceptibility. Airborne particulate trace metals are considered
to represent a health hazard since they can be
absorbed into human lung tissues during breathing
(Finlayson-Pitts and Pitts, 2000; Quiterio et al.,
2004a, b). A particular fraction of the particulate
matter (PM) which is known to exert toxic effects is
metals, such as Fe, Zn, and Cu, which may release
free radicals in lung fluid via the Fenton reaction,
and are hypothesized to cause cellular inflammation
(Birmili et al., 2006). Heavy metals are present in the
atmosphere in ever increasing levels as a result of
anthropic and natural emissions (Suzuki, 2006).
Frequently, anthropic emissions cause the levels of
metal in suspended particles to be well above
natural background levels (Finlayson-Pitts and
Pitts, 2000; Quiterio et al., 2004a, b).
Fossil fuel and wood combustion as well as waste
incineration and industrial processes are the main
anthropic sources of metals to the atmosphere. In
urban areas, vehicular emissions—and dust resuspension associated to road traffic—become the most
important manmade source (Birmili et al., 2006;
Preciado and Li, 2006; Sternbeck et al., 2002; Lin
et al., 2005; Harrison et al., 2003). It is believed that
burning of fossil fuels is responsible for Be, Co, Hg,
Mo, Ni, Sb, Se, Sn, V, Cr, Cu, Mn and Zn contents,
of PM associated with atmospheric particles (Allen
et al., 2001; Swaine, 2000), while vehicle ageing and
wearing actions release off some heavy metals such
as Zn from tires, Cu from brake linings, Mn from
moving metal parts and gasoline additives (Preciado
and Li, 2006; Swaine, 2000) and Pt, Pd and Rh from
catalytic converters of automobiles (Lesniewska
et al., 2006). Moreover, there are also indications
that heavy-duty vehicles are strong emitters of Ba
and Sb, but not of Cu, than light-duty vehicles
(Sternbeck et al., 2002). On the other hand,
metallurgical processes produce the largest emissions of As, Cd, Cu, Mn and Zn (Allen et al., 2001;
Swaine, 2000).
Natural emissions result from a variety of
processes acting on crustal minerals, including
volcanism, erosion and surface winds, as well as
from forest fires and the oceans (Allen et al., 2001;
Desboeufs et al., 2005). Even though estimations for
natural sources are difficult to be made, on a global
scale, resuspended surface dusts make a large
contribution to the total natural emission of trace
metals to the atmosphere, accounting for 450% of
Cr, Mn and V and 420% of Cu, Mo, Ni, Pb, Sb
and Zn while volcanic activity probably generates
up to 20% of atmospheric Cd, Hg, Cr, As, Cu, Ni,
Pb and Sb. Sea salt aerosols generated by spray and
wave action may contribute with around 10% of
total trace metal emissions, while elements contained in biological aerosols are important in
forested regions. Cu, Pb and Zn are contained in
emissions from biomass combustion (Allen et al.,
2001 and references therein; Desboeufs et al., 2005;
Swaine, 2000).
In a previous work (de Andrade et al., 1996), we
studied the metal concentrations in Salvador, Bahia,
Brazil in two sites: a residential area and a
residential–commercial–industrial neighborhood.
The determined metals associated to total suspended particles include Al, Ca, Cr, Cd, Fe, Mg,
Mn, V, Zn and Na. The result of factor analysis
suggested that the major primary sources in
Salvador are: resuspended soil, marine aerosol,
vehicular emission, metallurgy and building construction. It was noted that the contribution of
natural sources (soil and marine aerosol) accounted
for about 50% of the total deposition.
The aim of this work was to determine the
atmospheric concentrations and profiles of elements
such as iron, manganese, copper, and zinc, present
in each site, namely: (i) a bus station, strongly
impacted by heavy-duty diesel vehicles, and which
ARTICLE IN PRESS
P.A. de P. Pereira et al. / Atmospheric Environment 41 (2007) 7837–7850
has been already characterized for high molecular
weight n-alkanes and polycyclic aromatic hydrocarbons (de P. Pereira et al., 2002, 2003); (ii) a
harbor, impacted by an intense movement of goods,
including metal ores and concentrates and near
industrial centers and; (iii) an island, a non vehicleinfluenced site, with activities such as handcraft
work and fishery, although placed near the port.
2. Experimental
2.1. Collecting sites
The Lapa Bus Station is placed in downtown
Salvador City (81300 0000 S and 371300 0000 W), in a
region featured by heavy commerce and service
activities, with several stores, small office buildings
and a large mall at its neighborhood. The station is
composed of three floors: the first one being located
at underground level, and the second at the ground
level, with total areas of about 13,920 m2, each one
has five platforms for arrivals and departures of
urban heavy-duty diesel buses. Besides the movements of arrival and departure, the vehicles that are
waiting for departure remain with their motors on.
Finally, the third floor is occupied by small stores,
cafeterias and the administration services. The
station makes the confluence for several bus lines
coming from all other districts. Samples (n ¼ 35) of
PM10 particles (dpp10 mm) were collected between
16 and 28 July 2005 on quartz filters (22.8 17.7 cm,
Energética, RJ, Brazil), by placing a PM10 HI-VOL
sampler (Energética, RJ, Brazil and Thermo
Andersen, USA) on the ground level at the underground floor of the bus station, at an average flow
rate of 1.14 m3 min1. Sampling periods along
weekdays were 4–6 h during the morning/afternoon
and 8–10 h during the night, while on weekends they
were 24 h in order to cover—and to be able to
compare—rush-to-non-rush and weekday-to-weekend periods.
The Port of Aratu is placed on the Caboto harbor
(121470 0000 S and 131300 0000 W), inside the Todos os
Santos Bay, Candeias county, 50 km from Salvador
(Fig. 1). Its average temperature is 26 1C and the
predominant winds are NE (3–37%), E (3–52%)
and SE (3–50%). Nowadays, the Port of Aratu is
responsible for up to 60% of the total operations in
ports of the Salvador basin, giving support to the
industrial center of Aratu (CIA) and also to the
petrochemical complex of Camac- ari. It has four
terminals: one for gas, one for liquid and two for
7839
solid granaries. The solid granaries received by the
Port of Aratu include fertilizers, copper concentrates, manganese ores, mineral coke and phosphate
rocks. It is characterized by a very low traffic of
light- and heavy-duty vehicles, but an intense
movement of charge and discharge of solids,
to and from ships (http://www.codeba.com.br/
porto_aratu). In this site, particles were sampled
for both TSP (dpX100 mm) and PM10 fractions, by
placing the HI-VOL samplers together on the ground
level and operating them simultaneously, at average
flow rates of 1.16 and 1.14 m3 min1, respectively, in
two different points: firstly the Aratu#1 (7 samples),
near an office building and secondly the Aratu#2
(8 samples), in an open place. The samples were
collected from October/2004 to November/2004, in
sampling periods of 24 h each one.
Bananeira, Maré Island, is located on the Todos os
Santos Bay, about 300 m away from the Port of
Aratu, in a north–northwest direction (Fig. 1). It is
a ‘‘boat-only access’’ place and non vehicle-influenced village composed by 1000 inhabitants whose
activities are essentially based on handcraft work
and fishery. In this place, depending on the wind
direction, emissions may be coming from the port or
from industrial sites, as CIA and Petrochemical
Complex of Camac- ari. In this site, only samples of
TSP particles were collected, by placing the TSP HIVOL sampler on the roof of the local radio station
and operating it at an average flow rate of
1.16 m3 min1. Fourteen samples were collected
between September and October 2005, in 24 h
periods due to the low particle concentrations.
After each sampling, the filters were folded and
wrapped in aluminum foils and put inside sealed
plastic bags afterwards until weighing. Sample
masses were determined by weighing the filters before
and after sampling, using an analytical balance
(Sartorius Analytic, Goettingen, Germany). According to the standard procedure, before weighing, the
blank and sampled filters should be equilibrated for
24 h at a constant relative humidity lower than
5075% and at a constant temperature between 15
and 30 1C (within73 1C), by placing them inside a
desiccator (Method IO-3.1, 1999). After weighing,
the filters were folded and wrapped in aluminum foils
again, put inside sealed plastic bags and stored in the
lab in a refrigerator (4 1C) until analysis.
Summing up, while in the Port of Aratu both TSP
and PM10 samples were collected, in the Lapa
Station only PM10 and in the Bananeira only TSP
particles were sampled, in different sampling times
ARTICLE IN PRESS
7840
P.A. de P. Pereira et al. / Atmospheric Environment 41 (2007) 7837–7850
Copper
Complex
DOS FRADES
ISLAND
Caboto
MARE
ISLAND
Port of Aratu
ARATU BAY
Parafuso
Road
Sibra
ATLANTIC
OCEAN
TODOS OS
SANTOS BAY
Airport
Ferry Boat
ITAPARICA
ISLAND
Port of
Salvador
Site
Lapa
CABOTO
TODOS OS
SANTOS BAY
MARE ISLAND
Site
Bananeira
Site
Aratu #1 and # 2
Aratu Port
ARATU BAY
Fig. 1. Map localizing the three sites used in this study for collecting particulate samples.
for each place. The decision to collect samples of
different particle sizes was based on two factors:
(i) if a site was or was not supposed to be heavily
impacted by important sources of particulate heavy
metals, what should reflect on high or low levels of
metal contents on samples and (ii) the adjusted
availability of samplers to be used during sampling,
always keeping the first aspect as target due to its
importance to build a reliable sampling strategy in
order to collect representative samples of each local.
2.2. Sample preparation and analysis
A 47 mm diameter section of each filter
(17.34 cm2) was cut and used for the metal analysis,
according to the procedure described below.
The filters were extracted by adding 5 mL of nitric
acid (Merck Suprapur 65%), 2 mL of hydrochloric
acid (Merck Suprapur 36%) and 10 mL of ultrapure
water in a pyrex glass tube and let still for 2 h at
95 1C in a heating plate (Fernández et al., 2000). The
extracted solution was filtered, using a Whatman#41 (WH1441-110) filter, completed to 50 mL
with ultrapure water and kept in pre-cleaned
polyethylene bottles in the refrigerator until analysis
(Serrano et al., 1996; Beceiro-González et al., 1997).
A similar methodology was previously used for TSP
(Quiterio et al., 2004a, b).
Filter and reagent blanks were processed with the
same treatment. Metals were determined by ICP
OES (Perkin-Elmer, Optima 3000), according to
the Method IO-3.4 (1999). An external standard
ARTICLE IN PRESS
P.A. de P. Pereira et al. / Atmospheric Environment 41 (2007) 7837–7850
calibration curve was done for the following
elements: Mn, Fe, Zn, Cu, Co, Ni, Cd, Cr and Pb.
The content of the blanks for these nine
elements was o8% of the average content for
samples.
Both limit of detection (LOD) and accuracy
for the method were determined following Method
IO-3.4 (1999). LOD were computed as three times
the standard deviation of the distribution of outputs
for 10 repeated measurements of the standard,
which contained no metals (Boss and Fredeen,
1999). These LOD were calculated as 1.5 ng m3 for
Mn; 0.4 ng m3 for Cr, Cd, Ni; 0.5 ng m3 for Cu,
Co, 5 ng m3 for Fe, 1 ng m3 for Zn and Pb. The
accuracy of the method was evaluated using a
standard reference material (111355 ICP multielement Standard Solution IV from Merck) (Merck,
2006). The obtained results were in the acceptable
range of 3–8%. All samples and SRM were
determined in triplicate and a difference lower than
1% was considered acceptable.
7841
2.3. Statistical tests
Experimental data were analyzed by calculating
the Spearman’s correlation coefficients using STATISTICA 6.0 (Stat soft) program. Also Principal
Component Analysis (PCA) and Cluster Analysis
(CA), using Ward’s Method and Euclidian distances, were performed. The calculations were
performed using the individual experimental values
for each sample.
3. Results and discussion
3.1. TSP and PM10 particles in air and statistics
Table 1 shows a statistical summary of metal
(ng m3), TSP and PM10 (mg m3) concentrations
for Port of Aratu, Bananeira and Lapa Bus Station.
In spite the fact that nine different elements were
investigated (Cu, Fe, Mn, Zn, Cr, Co, Ni, Cd
and Pb), only Cu, Fe, Mn and Zn were found to be
Table 1
Statistical summary of elements, TSP and PM10 concentrations in Aratu, Bananeira and Lapa Bus Station
Elementa (ng m3)
Site
Cu
Aratu#1 TSP
Mean
SD
Min
Max
Aratu#2 TSP
Mean
SD
Min
Max
Aratu#1 PM10
Mean
SD
Min
Max
Aratu#2 PM10
Mean
SD
Min
Max
Bananeira TSP
Mean
SD
Min
Max
3.37
2.54
0.49
8.57
Lapa Bus Station PM10
Mean
SD
Min
Max
1.87
1.04
0.59
5.88
a
21.0
14.3
7.16
47.6
121
91.9
42.1
293
8.79
3.30
5.19
13.2
79.2
82.8
17.8
257
Particulate matter (mg m3)
Mn
226
211
48.6
596
19.2
11.3
5.92
39.2
4.53
2.82
1.41
9.40
182
87.7
106
354
7
328
199
93.3
650
13.5
4.70
5.14
20.0
3.95
1.26
2.23
6.35
169
46.4
95.9
222.6
8
79.6
60.1
18.5
165
8.32
5.81
2.47
16.8
1.83
0.33
1.43
2.20
64.6
30.1
44.2
123
7
205
119
75.0
470
9.52
6.36
2.61
21.2
2.80
0.72
1.96
3.69
71.7
13.9
49.2
88.2
8
26.8
17.6
7.03
75.0
149
72.9
52.0
343
19.5
22.3
1.60
81.8
9.68
9.22
1.81
44.9
Zn
145
56.9
68.0
243
TSP
PM10
n
Fe
36.1
10.1
16.9
54.5
5.78
7.31
1.49
40.8
Elements with mean concentration below their LOD were not included (Cr, Co, Ni, Cd and Pb).
15
112
68.9
30.9
393
35
ARTICLE IN PRESS
7842
P.A. de P. Pereira et al. / Atmospheric Environment 41 (2007) 7837–7850
above their limits of detection, evidencing that those
studied sites have no important sources of Cr, Co,
Ni, Cd and Pb until the present. These five elements
were disconsidered in all samples collected in the
four locations (Aratu#1 n ¼ 7 for TSP and PM10;
Aratu#2 n ¼ 8 for TSP and PM10; Bananeira n ¼ 15
and Lapa Station n ¼ 35) because individual
determinations and the arithmetic mean concentrations were below the LOD. Even though keeping in
mind that particulate matter of different size
fractions were collected on each site and by
considering only arithmetic means from Table 1, it
was found that Fe was the element in the highest
concentrations in all sites, except for Bananeira,
where Zn presented the highest concentration.
Because Port of Aratu was the unique site where
both TSP and PM10 samples were simultaneously
collected in two different places, the F-test was
performed in order to find whether these factors
could be affecting the determined concentrations of
the samples. It was found (Table 2) that neither the
two chosen places for collection nor TSP and PM10
fractions should be considered as ‘‘similar’’ for
following comparisons. Although they were not so
far from each other, Aratu#1 was near a building,
while Aratu#2 was in an open place. This geographical peculiarity could contribute, for example, to
differences in the wind’s predominant direction and
velocity that each one would receive. Also, and
maybe more important, the samples in the two
places were collected in different periods. By
consequence throughout this paper, samples from
Aratu#1 and Aratu#2 were computed separately,
not trying to compare them in any way. On the
other hand, since TSP and PM10 samples were
collected simultaneously in each of these two sites,
they could be compared when it would be necessary.
The PM10 concentrations, as function of the
sampling period in the Lapa station, are shown in
Fig. 2. The highest concentrations were generally
found in the afternoon, since the rush hours—when
there were an increase in the frequency of arrivals
Table 2
F test for analyzing PM10 and TSP fractions from Port of Aratu
(95% of confidence limit)
Analyzes
Aratu#1
Aratu#1
Aratu#1
Aratu#2
Fcalc
(TSP) vs. Aratu#2 (TSP)
3.57
(PM10) vs. Aratu#2 (PM10) 4.69
(TSP) vs. Aratu#1 (PM10)
8.49
(TSP) vs. Aratu#2 (PM10) 11.1
Fcrit
Conclusion
3.87
3.97
4.95
3.79
Similar
Non-similar
Non-similar
Non-similar
and departures of buses—were covered during this
sampling period. Since an increase in commuting
could promote high concentrations of particulate
matter, coming from diesel burning in heavy-duty
vehicles and particles resuspension associated to the
local traffic, it seems reasonable that afternoon
period has shown higher levels of the determined
elements and PM10. The lowest levels, on the
contrary, were found at night due to a reduction
in the number of vehicles circulating in the station
and/or in particles resuspension. This trend has been
already observed with polycyclic aromatic hydrocarbons (PAH) concentrations, in samples collected
in 1991 at the same place. (de P. Pereira et al., 2002).
It should be highlighted that almost all samples
(91%) presented concentrations higher than the
50 mg m3 Brazilian primary standard for PM10,
while six (17%) were higher than the secondary
standard of 150 mg m3 for 24 h exposure.
A F-test was performed in order to find the
statistical significance of differences in concentrations
reported in Fig. 2. Data were compared in pairs:
morning vs. afternoon (Fcalc ¼ 0.37), morning vs.
night (Fcalc ¼ 0.37) and afternoon vs. night (Fcalc ¼
0.99). In the three situations, the critic F-value was
0.26, indicating that data are non-similar.
Pearson’s correlation (Table 3) shows low-tomoderate correlations among studied metals for
Lapa Station, while the PCA analysis (which
explains only 59% of results) shows that Fe, Cu,
Mn and Zn form only one component (Table 4). In
the Cluster Analysis (Fig. 3), a sub-group is formed
between Zn and Cu and a second, with a more
strong linkage, between Fe and Mn, and both
associating with each other. Also, the stronger
correlation among metal concentrations was found
for Mn and Fe (r ¼ 0.78). This could be indicative
of soil resuspension associated to buses traffic
as the main sources responsible for those metal
concentrations.
TSP and PM10 average concentrations in Aratu
were higher than the maximum allowable by the
Brazilian standards for annual values. For TSP,
they were in Aratu#1 and Aratu#2, respectively, 182
and 169 mg m3, while for PM10 they were 64.6 and
71.7 mg m3 in the same sites. Both Lapa Station
and Port of Aratu show a not so much different
profile for the metals determined, with iron being
the most abundant one (Table 1, Figs. 2 and 4).
Average concentrations for iron in PM10 filters
were, respectively, 149, 79.6 and 205 ng m3 for
Lapa, Aratu#1 and Aratu#2. Despite these high
ARTICLE IN PRESS
P.A. de P. Pereira et al. / Atmospheric Environment 41 (2007) 7837–7850
7843
180.00
n=35
Lapa (Response)
150.00
120.00
90.00
60.00
30.00
0.00
PM10 (ug m-3)
Cu (ng m-3)
Fe (ng m-3)
Mn (ng m-3)
Zn (ng m-3)
200.0
morning (n=12)
afternoon (n=12)
Lapa (Response)
160.0
night (n=11)
120.0
80.0
40.0
0.0
PM10 (ug m-3)
Cu (ng m-3)
Fe (ng m-3)
Mn (ng m-3)
Zn (ng m-3)
Fig. 2. (a) Arithmetic mean values for PM10 and metal concentration for Lapa Bus Station and (b) morning-to-afternoon-to-night
comparisons for PM10 and metal concentrations based on arithmetic means for Lapa Bus Station. In both cases, bars indicate7standard
errors.
values, they were usually lower than those found,
for example, in seven districts of the Baixada
Fluminense, in Rio de Janeiro, where pollution
sources are mainly local industries and vehicular
emissions (Quiterio et al., 2005).
The Pearson’s statistics for Aratu#1 TSP and
Aratu#2 TSP shows that correlations are slightly
better only for Aratu#1, while in Aratu#2 metals
were not correlated, except for Cu and Mn
(Table 3). By Cluster Analysis it is shown a more
clear correlation between Cu and Mn, while Fe is
located in another group, giving support to the
hypothesis of Fe is coming mainly from soil
resuspension, for both sites (Fig. 3). These results
are consistent with low correlations between Fe and
Cu (0.05 and 0.48, respectively) and Fe and Mn
(0.04 and 0.54, respectively). Aratu#1 and
Aratu#2 PM10 presented higher scores for Pearson’s
correlations when compared to TSP samples. This
fact may be explained by the different contributions
ARTICLE IN PRESS
P.A. de P. Pereira et al. / Atmospheric Environment 41 (2007) 7837–7850
7844
to fine and coarse particulate matter composition.
The Cluster analysis shows Zn, Mn and Cu together
in one group and Fe in the other one. Discharge of
solid granaries, as fertilizers and metal ores, in the
port’s terminal may be the main sources for Cu, Mn
Table 3
Pearson’s correlations of metals from all sites (significant values
at 95% of confidence limit are in bold)
Cu
Lapa Station
Cu
Fe
Mn
Zn
Fe
Mn
Zn
1.0
0.35
0.45
0.18
1.0
0.78
0.52
1.0
0.37
1.0
Aratu#1 TSP
Cu
1.0
Fe
0.05
Mn
0.19
Zn
0.57
1.0
0.04
0.57
1.0
0.32
1.0
Aratu#1 PM10
Cu
Fe
Mn
Zn
1.0
0.15
0.48
0.98
1.0
0.41
0.19
1.0
0.57
1.0
1.0
0.33
0.07
0.33
1.0
0.75
0.30
1.0
0.46
1.0
Aratu#2 TSP
Cu
1.0
Fe
0.48
Mn
0.52
Zn
0.71
1.0
0.54
0.07
1.0
0.05
1.0
Aratu#2 PM10
Cu
1.0
Fe
-0.25
Mn
0.82
Zn
0.21
1.0
0.16
0.51
1.0
0.41
1.0
Bananeira
Cu
Fe
Mn
Zn
and Zn while, as previously stated, Fe may be due to
soil resuspension.
The receptor site of Bananeira, on the other hand,
has presented a different profile when compared to
the other sites. Primarily, the mean TSP concentration (36.1 mg m3) was, as expected due to characteristics of the site, not only much lower than the
other two places studied, but also lower than values
reported for two urban districts in Salvador and
localities in Rio de Janeiro and other cities (de
Andrade et al., 1996). The stronger correlation
among metals was found for Mn and Fe (r ¼ 0.75)
indicating that soil resuspension may be the main
source of these metals.
In terms of the elements determined (Fig. 5), zinc
was the most abundant, presenting average concentration of 145 ng m3 and ranging from 68 to
243 ng m3, measured in TSP samples. Iron, instead,
was comparatively low (26.8 ng m3 average), while
copper was lower than in Aratu but higher than in
the Lapa Station.
Zinc, to our present knowledge, has no significant
source contributions coming from the Port of
Aratu. Thus, its high concentrations strongly
suggest the hypothesis that there are other probable
sources impacting the Bananeira site, besides Aratu,
and these could be the Industrial Center of Aratu
and/or the siderurgy plant, not too distant from the
location.
3.2. Enrichment factors
Trace metals in aerosols are derived from a
variety of sources which include the Earth’s crust,
the oceans, volcanic activity, the biosphere, and a
number of anthropogenic processes (e.g., fossil fuel
burning, waste incineration, the processing of ores,
etc.). The degree to which a trace metal in an aerosol
is enriched, or depleted, relative to a specific source
can be assessed to a first approximation using an
enrichment factor (EF). It has been the practice to
Table 4
Principal Component Analysis (at 95% of confidence limit)
Cu
Fe
Mn
Zn
Lapa
Bananeira
PC1
PC1
0.57
0.89
0.87
0.69
0.01
0.85
0.92
0.67
Aratu#1 TSP
Aratu#2 TSP
Aratu#1 PM10
Aratu#2 PM10
PC2
PC1
PC2
PC1
PC2
PC1
PC2
PC1
PC2
0.92
0.37
0.06
0.55
0.67
0.63
0.34
0.96
0.12
0.59
0.82
0.01
0.93
0.68
0.74
0.55
0.28
0.57
0.44
0.82
0.93
0.20
0.68
0.99
0.06
0.93
0.41
0.03
0.92
0.10
0.96
0.52
0.32
0.93
0.001
0.75
ARTICLE IN PRESS
1400
700
1200
600
1000
500
Linkage Distance
Linkage Distance
P.A. de P. Pereira et al. / Atmospheric Environment 41 (2007) 7837–7850
800
600
400
200
300
200
0
Fe
Mn
Zn
Cu
Fe
1200
1400
1000
1200
Linkage Distance
Linkage Distance
400
100
0
800
600
400
200
Mn
Zn
Cu
1000
800
600
400
200
0
0
Fe
Mn
Zn
Cu
Fe
350
900
300
800
Mn
Zn
Cu
700
250
Linkage Distance
Linkage Distance
7845
200
150
100
600
500
400
300
200
50
100
0
0
Fe
Mn
Zn
Cu
Fe
Mn
Zn
Cu
Fig. 3. Cluster analyses at 95% of confidence limit: (a) Lapa’s Station; (b) Bananeira; (c) Aratu#1 TSP; (d) Aratu#2 TSP; (e) Aratu#1
PM10; (f) Aratu#2 PM10.
categorise trace metals in the aerosol on the basis of
their crustal EFs, calculated according to Al or Fe
contents as follows:
EFcrust ¼ ðC xp =C Fep Þ=ðC xc =C Fec Þ,
(1)
where (Cxp) and (CFep) are the concentrations of a
trace metal X and Fe (or Al) (which are used as
crustal reference element), respectively, in the
aerosol, and (Cxc) and (CFec) are their concentra-
tions in average crustal material. By convention, an
arbitrary average EF value of o10 is taken as an
indication that a crust trace metal in an aerosol has
a significant crustal source, and these are termed the
non-enriched elements (NEEs). In contrast, an EF
value of 410 crust is considered to indicate that a
significant proportion of an element has a noncrustal source, and these are referred to the
anomalously enriched elements (AEEs); however,
ARTICLE IN PRESS
P.A. de P. Pereira et al. / Atmospheric Environment 41 (2007) 7837–7850
7846
500.0
PM10 (n=6)
PTS (n=8)
Aratu#1 (Response)
400.0
300.0
200.0
100.0
0.0
Gravimetric Cu (ng m-3)
(ug m-3)
Fe (ng m-3)
Mn (ng m-3)
Zn (ng m-3)
350.0
PM10 (n=6)
PTS (n=8)
Aratu#2 (Response)
300.0
250.0
200.0
150.0
100.0
50.0
0.0
Gravimetric Cu (ng m-3)
(ug m-3)
Fe (ng m-3)
Mn (ng m-3)
Zn (ng m-3)
Fig. 4. Arithmetic mean values for TPS, PM10 and metal concentrations for (a) Aratu#1 and (b) Aratu#2 sites. Bars indicate7standard
errors.
when sufficient crustal material is present in the air
the AEEs can switch character and behave as NEEs.
This is an essentially crude classification, but is
useful in interpreting aerosol chemistry (Caroli
et al., 1996; Ure and Davidson, 1995; Wedepohl,
1995; Chester et al., 1999; Odabasi et al., 2002;
Herut et al., 2001; Sardans and Peñuelas, 2006).
EFs in this study (Table 5) were calculated based
on Fe natural content, since due to previously
assumed storage protocols for determination of
the organic species, in the same filters, aluminum
foils were used to wrap the filters before and after
sampling and thus that element could not be
analyzed.
When EFs based in the iron concentration were
calculated for the Lapa samples, high values were
found for zinc and copper, 28 and 19, respectively,
and low for manganese (4). This low EF for
manganese, together with its high Pearson’s correlation with iron (0.78) suggests that soil resuspension
is the unique or the main source of that element.
The high EF values for zinc and copper, on the
other hand, could denote them as AEEs although
Pearson’s correlation between Zn and Fe (0.52) and
Cu and Fe (0.35) support the hypothesis that soil
resuspension by buses circulation is also an important source for both elements in the Lapa site. In
spite of this fact, previous works (Pacyna 1986;
ARTICLE IN PRESS
P.A. de P. Pereira et al. / Atmospheric Environment 41 (2007) 7837–7850
7847
150.00
Bananeira (Response)
120.00
90.00
60.00
30.00
0.00
PTS (ug m-3) Cu (ng m-3)
Fe (ng m-3)
Mn (ng m-3)
Zn (ng m-3)
Fig. 5. Arithmetic mean values for TPS and metal concentrations for Bananeira’s site. Bars indicate7standard errors.
Table 5
Enrichment Factors based on arithmetic means and relative to Fe for all sites (Enrichment Factors based on arithmetic means and relative
to Fe for all sites (values in bold indicate anthropogenic contributions as the main source)
Cu
Fe
Mn
Zn
Lapa
Bananeira
Aratu#1 TSP
Aratu#1 PM10
Aratu#2 TSP
Aratu#2 PM10
19
1.0
4.0
28
192
1.0
45
3909
142
1.0
5.2
15
168
1.0
6.4
17
564
1.0
2.5
8.7
589
1.0
2.9
9.9
López et al., 2005) have reported copper as a
component in emissions from both gasoline and
diesel fueled vehicles and in wearing of brakes, and
zinc as a component of tire rubber debris. Furthermore, the low correlation between these two
elements (0.18) suggests yet other sources for them.
Besides the larger amounts of iron than Aratu#1,
Aratu#2 had also an increase in the copper
concentration and in the EF of this element (Tables
1 and 5 and Fig. 4). Looking to the EFs for copper
they were, respectively, 142 and 564 for TSP and
168 and 589 for PM10, showing the relative increase
of the copper concentration and EF in Aratu#2 and
denoting this element as an AEE. Firstly, these high
concentrations and EFs of copper may probably be
due to the port’s activities, since discharges of
copper concentrate are frequent at its terminal.
Also, other possible sources would be the burning,
in ships, of diesel fuel and heavy oil in heaters. The
low correlations shown between this element and
iron (r ¼ 0.05 and 0.48 for TSP and 0.15 and
0.25 for PM10 for Aratu#1 and #2, respectively)
reinforces the hypothesis of other sources for it,
besides soil resuspension. Thus, soil resuspension
does not seem to be the predominant source for
these two locations. Actually, copper content in PM
mostly reflects both port’s activities and burning of
diesel and heavy oil on the terminal as main sources.
The EF for zinc in Bananeira was very high
(3909), showing strong anthropic contributions, and
this was followed by copper (192) and manganese
(45). Because of its unexpected high EF and that
there are no significant sources of this element
neither on Aratu Harbor nor on Bananeira, other
very significant emission sources of zinc seem to be
acting on this scenery. They probably would be the
Industrial Center of Aratu (CIA) and/or the Sibra/
RDM siderurgy plant, not too distant from the
island.
It should be highlighted that while copper had
also a high EF (192), its correlation with zinc was
poor (0.33), what suggests different sources for
both elements. Manganese, despite its moderately
high EF (45), had a good correlation with iron
(0.75) and its main source could be the soil
resuspension.
ARTICLE IN PRESS
P.A. de P. Pereira et al. / Atmospheric Environment 41 (2007) 7837–7850
7848
3.3. Estimate of dry deposition fluxes
The dry atmospheric deposition fluxes (Fd) were
calculated by multiplying the geometric mean
particulate concentration in air of the element of
interest (i) by the elemental dry settling velocity
(Vd):
F d ¼ Ci V d.
(2)
The term Vd varies with particle size and is
dependent on climatological and physical conditions in the troposphere, especially in coastal
environments (Herut et al., 2001). The values used
here follow the mean values used by Duce et al.
(1991). For Zn and Cu a mean value of 0.1 cm s1
was applied, and for Fe and Mn a mean value of
2 cm s1 was adopted. These values fall close to the
Vd range given in other studies (Rojas et al., 1993;
Migon et al., 1997; Herut et al., 2001 and references
therein). It should be emphasized, however, that the
Table 6
Estimates of trace element dry deposition fluxes (Fd)
Cu Fe
a
1
Mn
2.0
Zn
Dry settling velocity (Vd) (cm s )
0.1
2.0
0.1
Lapa
Geometric Mean (ng m3)
Dry deposition Flux (Fd)
(mg m2 yr1)
1.6 133.5 7.1
0.4 674
36
4.2
1
Bananeira
Geometric Mean (ng m3)
Dry deposition Flux (Fd)
(mg m2 yr1)
2.5 22
2
338
11
159
134
101
Aratu#1 TSP
Geometric Mean (ng m3)
Dry deposition Flux (Fd)
(mg m2 yr1)
18
13
160
2416
16
243
4
3
Aratu#1 PM10
Geometric Mean (ng m3)
Dry deposition Flux (Fd)
(mg m2 yr1)
8
6
61
925
7
102
2
1
Aratu#2 TSP
Geometric Mean (ng m3)
Dry deposition Flux (Fd)
(mg m2 yr1)
96
73
269
4072
13
190
4
3
Aratu#2 PM10
Geometric Mean (ng m3)
Dry deposition Flux (Fd)
(mg m2 yr1)
52
40
181
2742
8
117
3
2
a
flux calculations might vary by approximately an
order of magnitude due to the uncertainties in Vd.
Table 6 shows dry deposition fluxes for Cu, Fe,
Mn and Zn content of particulate matter from Lapa
Bus Station, Port of Aratu and Bananeira. Fe is the
element which demonstrated higher dry deposition
flux, followed by Mn, Cu and Zn.
According to Herut et al. (2001).
4. Conclusions
In this work, the atmospheric concentrations and
profiles for iron, manganese, copper and zinc were
determined in three different sites, namely: (i) a bus
station, strongly impacted by heavy-duty diesel
vehicles, (ii) a harbor, impacted by an intense
movement of goods, including metal ores and
concentrates and near industrial centers and; (iii)
an island, a non-vehicle-influenced site, with activities such as handcraft work and fishery. Metals
were determined by ICP OES with external standard calibration curves. The average concentrations
for PM10 and TSP, in the bus station and in the
Aratu harbor, were higher than the values allowable
by the Brazilian standards, while for the Bananeira
site it was lower, as expected due to the local
characteristics.
The main sources for iron in the Lapa Station and
Aratu harbor were, respectively, the soil resuspension by buses and discharge of solid granaries, as
fertilizers and metal ores.
Zinc and copper in the Lapa station were mainly
from anthropic contributions as, for example,
components in emissions from both gasoline and
diesel fueled vehicles, wearing of brakes and tire
rubber debris. Nevertheless, the soil resuspension
provoked by buses movement has also contributed.
In the Aratu site, copper predominant sources
were either the port’s activities such as discharges of
copper concentrates on its terminal or burning of
diesel fuel and heavy oil in heaters in ships although
other minor sources may be acting as well.
The Bananeira site has presented a different
profile, with zinc as the most abundant element,
with concentrations ranging from 68 to 243 ng m3,
measured in TSP samples. The EF for zinc was very
high (3909), showing strong anthropic contributions, followed by copper (192) and manganese (45).
Zinc, to our present knowledge, has no significant
source contributions coming from the Port of
Aratu, and its high concentrations strongly suggest
the hypothesis of other probable sources, as the
ARTICLE IN PRESS
P.A. de P. Pereira et al. / Atmospheric Environment 41 (2007) 7837–7850
Industrial Center of Aratu and/or the siderurgy
plant, not too distant from the location.
Copper had also a high EF (192), but its
correlation with zinc was poor (0.33), what
suggests different sources for both elements. Manganese, despite its moderately high EF (45), had a
good correlation with iron (0.75) and its main
source could be the soil resuspension.
Finally, in the Aratu samples, iron, manganese
and copper were present predominantly in the PM10
fraction of the particulate matter, instead of TSP,
what might be contributing to health risks to the
local workers.
Acknowledgments
The authors wish to thank to Conselho Nacional de
Desenvolvimento Cientı́fico e Tecnológico (CNPq),
Fundac- ão de Amparo a Pesquisa do Estado da Bahia
(FAPESB), Financiadora de Estudos e Projetos
(FINEP), Agência Nacional de Energia Elétrica
(ANEEL), Nordeste Generation, RECOMBIO/
CNPq/CTENERG and to Eng. Oscar A. Reis
Martinez, for helpful discussions.
References
Allen, A.G., Nemitz, E., Shi, J.P., Harrison, R.M., Greenwood, J.C.,
2001. Size distributions of trace metals in atmospheric
aerosols in the United Kingdom. Atmospheric Environment
35, 4581–4591.
de Andrade, J.B., Macedo, M.A., Korn, M., Oliveira, E.,
Gennari, R.F., 1996. A comparison study of aerosol emissions
sources in two receptor sites in Salvador (Brazil) city.
Technological and Environmental Chemistry 54, 23–28.
Beceiro-González, E., Andrade-Garda, J.M., Serrano-Velasco, E.,
Lópes-Mahı́a, P., 1997. Metals in airborne particulate matter
in La Coruña (NW Spain). Science of the Total Environment
196, 131–139.
Birmili, W., Allen, A.G., Bary, F., Harrison, R.M., 2006. Trace
metal concentrations of water solubility in size-fractionated
atmospheric particles and influence of road traffic. Environmental Science and Technology 40, 1144–1153.
Boss, C.B., Fredeen, K.J., 1999. Concepts, Instrumentation, and
Techniques in Inductively Coupled Plasma Optical Emission
Spectrometry, second ed. Perkin-Elmer Corp., Norwalk.
Caroli, S., La Torre, F., Petrucci, F., Violante, N., 1996. Element
Speciation in Bioinorganic Chemistry. Wiley-Interscience,
New York.
Chester, R., Nimmo, M., Preston, M.R., 1999. The trace metal
chemistry of atmospheric dry deposion samples collected at
Cap Ferrat: a coastal site in the Western Mediterranean.
Marine Chemistry 68, 15–30.
Desboeufs, K.V., Sofitikis, A., Losno, R., Colin, J.L., Ausset, P.,
2005. Dissolution and solubility of trace metals from natural
7849
and anthropogenic aerosol particulate matter. Chemosphere
58, 195–203.
Duce, R.A., Liss, P.S., Merril, J.T., Atlas, E.L., Buat-Menard, P.,
Hickes, B.B., Miller, J.M., Prospero, J.M., Arimoto, R.,
Church, T.M., Ellis, W., Galloway, J.N., Hansen, L.,
Jickells, T.D., Knap, A.H., Reinhardt, K.H., Schneider, B.,
Soudine, A., Tokos, J.J., Tsunogai, S., Wollast, R., Zhou, M.,
1991. The atmospheric input of trace species to the world
ocean. Global Biogeochemical Cycles 5, 193–259.
Fernández, A.J., Ternero, M., Barragán, F.J., Jiménez, J.C.,
2000. An approach to characterization of sources of urban
airbone particles through heavy metal speciation. Chemosphere 2, 123–136.
Finlayson-Pitts, B.J., Pitts Jr., J.N., 2000. Chemistry of the Upper
and Lower Atmosphere: Theory, Experiments, and Applications, first ed. Academic Press, San Diego.
Harrison, R.M., Tilling, R., Romero, M.S.C., Harrad, S., Jarvis, K.,
2003. A study of trace metals and polycyclic aromatic
hydrocarbons in the roadside. Environmental Science and
Technology 37, 2391–2402.
Herut, B., Nimmo, M., Medway, A., Chester, R., Krom, M.D.,
2001. Dry atmospheric inputs of trace metals at the
Mediterranean coast of Israel (SE Mediterranean): sources
and fluxes. Atmospheric Environment 35, 803–813.
/http://www.codeba.com.br/porto_aratuS; accessed on June,
2006.
Lesniewska, B.A., Godlewska-Zylkiewicz, B., Ruszczynska, A.,
Bulska, E., Hulanicki, A., 2006. Elimination of interferences
in determination of platinum and palladium in environmental
samples by inductively coupled plasma mass spectrometry.
Analytica Chimica Acta 564, 236–242.
Lin, C.-C., Chen, S.-I., Huang, K.-L., Hwang, W.-I., ChangChien, G.-P., Lin, W.-Y., 2005. Characteristics of metals in
nano/ultrafine/fine/coarse particles collected beside a heavily
trafficked road. Environmental Science and Technology 39,
8113–8122.
López, J.M., Callén, M.S., Murillo, R., Garcı́a, T., Navarro,
M.V., de la Cruz, M.T., Mastral, A.M., 2005. Levels of
selected metals in ambient air PM10 in an urban site of
Zaragoza (Spain). Environmental Research 99, 58–67.
Merck, 2006. Retrieved June 05, 2006, from /http://chemdat.
merck.de/pls/pi03/web2.zoom_in?text=111355&screen=110&
cid=xcsfs&pg=0&s=&lang=15S.
Method IO-3.1, 1999. Selection, preparation and extraction of
filter material. US-EPA.
Method IO-3.4, 1999. Determination of metals in ambient
particulate matter using inductively coupled plasma (ICP)
spectroscopy. US-EPA.
Migon, C., Journel, B., Nicolas, E., 1997. Measurement of trace
metal wet, dry and total atmospheric fluxes over the Ligurian
Sea. Atmospheric Environment 31, 889–896.
Odabasi, M., Muezzinoglu, A., Bozlaker, A., 2002. Ambient
concentrations and dry deposition fluxes of trace elements in
Izmir, Turkey. Atmospheric Environment 36, 5841–5851.
Pacyna, J.M., 1986. In: Nriagu, J.O., Davidson, C.I. (Eds.), Toxic
Metals in the Atmosphere. Wiley, New York.
de P. Pereira, P.A., de Andrade, J.B., Miguel, A.H., 2002.
Measurements of semivolatile polycyclic aromatic hydrocarbons in a bus station and an urban tunnel of Salvador, Brazil.
Journal of Environmental Monitoring 4 (4), 558–561.
de P. Pereira, P.A., de Andrade, J.B.; Miguel, A.H., 2003. PAHs
and aliphatic hydrocarbons (C19 to C29) in a bus station and
ARTICLE IN PRESS
7850
P.A. de P. Pereira et al. / Atmospheric Environment 41 (2007) 7837–7850
an urban tunnel of Salvador, Bahia, Brazil. In: A & WMA’S
96TH Annual Conference & Exhibition, 2003, San Diego—
CA. Energy, Economic and Global Challenges—Environment in the Balance.
Preciado, H.F., Li, L.Y., 2006. Evaluation of metal loadings and
bioavailability in air, water and soil along two highways of
British Columbia, Canada. Water, Air, and Soil Pollution
172, 81–108.
Quiterio, S.L., Arbilla, G., Silva, C.R.S., Escaleira, V., 2004a.
Metals in airborne particulate matter in the Industrial District
of Santa Cruz, Rio de Janeiro, in an annual period.
Atmospheric Environment 38, 321–331.
Quiterio, S.L., Arbilla, G., Escaleira, V., Silva, C.R.S., Maia,
L.F.P.G., 2004b. Metals in airborne particulate matter in
downtown Rio de Janeiro (Brazil). Bulletin of Environmental
Contamination and Toxicology 72, 916–922.
Quiterio, S.L., Sousa, C.R., Arbilla, G., Escaleira, V., 2005.
Evaluation of levels, sources and distribution of airborne
trace metals in seven districts of the Baixada Fluminense,
Rio de Janeiro, Brazil. Atmospheric Environment 39,
3503–3512.
Rojas, C.M., Injuk, J., Van Grieken, R.E., Laane, R.W., 1993.
Dry and wet depostion fluyes of Cd, Cu, Pb and Zn into the
sourthern Bight of the North Sea. Atmospheric Environment
27A, 251–259.
Sardans, J., Peñuelas, J., 2006. Introduction of the factor of
portioning in the lithogenic enrichment of trace element
bioaccumulation in plant tissues. Environmental Monitoring
and Assessment 115, 473–498.
Serrano, E., Beceiro, E., López, P., Prada, D., 1996. Heavy
metals determination in atmospheric particulate matter of La
Coruña. Quimica Analitica 15, 38–44.
Sternbeck, J., Sjödin, A., Andréasson, K., 2002. Metal emissions
from road traffic and the influence of resuspension—results
from two tunnel studies. Atmospheric Environment 36,
4735–4744.
Suzuki, K., 2006. Characterization of airborne particulates and
associated trace metals deposited on tree bark by ICP-OES,
ICP-MS, SEM-EDX and laser ablation ICP-MS. Atmospheric Environment 40, 2626–2634.
Swaine, D.J., 2000. Why trace elements are important. Fuel
Processing Technology 65–66, 21–33.
Ure, A.M., Davidson, C.M., 1995. Chemical Speciation in the
Environment. Academic & Professional, London, Blackie, p. 408.
Wedepohl, K.H., 1995. The composition of the continental crust.
Geochimica et Cosmochimica Acta 59, 1217–1232.